Goal Exploration via Adaptive Skill Distribution for Goal-Conditioned Reinforcement Learning